# Statistical Decision Tree

> Interactive web application for selecting statistical tests and models in HCI research, teaching, and early analysis planning.

## Canonical URL

- https://valentin-schwind.github.io/statistics-decision-tree/

## Summary

- The Statistical Decision Tree is part of the HCI User Studies Toolkit.
- It guides users through study-design questions to identify suitable statistical tests or models.
- It includes assumption checks, effect-size guidance, post-hoc options, interpretation notes, and example code in R and Python.
- The main audience is students, novice researchers, teachers, and practitioners working with HCI or adjacent empirical user studies.

## Main Topics Covered

- Statistical test selection
- Parametric versus nonparametric routes
- t-tests, ANOVA, repeated-measures ANOVA, ANCOVA
- Linear regression and multiple regression
- Linear mixed models and robust alternatives
- Chi-square, Fisher's exact, McNemar, and marginal homogeneity tests
- Logistic regression, ordinal regression, count regression
- MANOVA, PERMANOVA, ART, GEE, and GLMM-related guidance
- Post-hoc procedures, effect sizes, and interpretation help

## Intended Use

- Teaching introductory and intermediate statistics for HCI and user studies
- Supporting early study planning and analysis orientation
- Helping users identify relevant model families before deeper statistical consultation

## Limitations

- The tool is didactic orientation support, not a substitute for statistical supervision.
- Final method choices still depend on study design, measurement quality, missingness, assumptions, and substantive research goals.

## Maintainer

- Prof. Dr. Valentin Schwind
- Hochschule der Medien Stuttgart
- https://valentin-schwind.de

## Source

- Repository: https://github.com/valentin-schwind/statistics-decision-tree
- Toolkit paper: https://doi.org/10.1145/3544549.3585890